TL;DR
Paid social is a creative-testing business dressed up as a media-buying business. The account structure, the bid strategy, and the attribution model matter less than the pace and quality of creative tests. The teams that win at paid social ship more angles per week than their competitors, kill losers fast, and let the platforms optimize inside a simple structure they rarely change.
The playbook, in one paragraph
Pick two platforms where the buyer is actually reachable. Build a simple account structure: one prospecting campaign per platform, one retargeting campaign, one branded-search or bottom-funnel campaign if the category supports it. Ship three to five new creative concepts per week per platform. Test at $50 to $150 per creative per day. Kill anything that does not hit a defined creative-quality threshold in the first 48 to 72 hours. Duplicate winners into scaling ad sets at 20 percent budget lifts every 48 hours. Feed the winners back into the creative team as a repeatable format. Measure with a triangulated model: platform-reported CPA, self-reported intake data, and quarterly incrementality tests. Kill any campaign that cannot defend itself when you turn it off.
Where this fits in the modern discovery layer
Paid social sits inside the paid discovery cluster of the nineteen Ranking Surfaces. It is a spend-driven surface, not a compounding organic one, but the way it is run interacts with several organic surfaces in ways that matter to the whole marketing system.
Paid Feed Optimization (PFO). The Meta feed, the TikTok For You Page, and the LinkedIn feed as paid ad units. This is the surface paid social competes on directly. Winning creative, correctly-structured audiences, and honest bidding are what make PFO produce revenue at a defensible cost.
Marketplace Surface Optimization (MSO). Reddit, Pinterest, and increasingly TikTok Shop function as intent marketplaces where paid placement drives commerce. Different creative language, different audience logic, but the same discipline of test-kill-scale.
Feedback into Social Search Optimization (SSO). Paid social is the fastest cheap signal on which organic creative angles will work. A boosted post that outperforms tells the organic team what to publish next week. Programs that treat paid and organic as separate teams give up this loop.
Feedback into Video Discovery Optimization (VDO). The paid-tested TikTok or Reels hook that broke out is usually the same hook that ranks in native platform search for months after the paid budget ends. Paid buys the read; VDO keeps the revenue.
What paid social is not: it is not brand marketing on its own, it is not a substitute for organic, and on most accounts it is not the highest-return channel. It is the fastest, most testable, and most learn-heavy channel, which is why it is often the first paid channel a team gets right.
The five levers
1. Creative testing pace
The single biggest predictor of paid social performance across accounts I have run is how many net-new creative concepts hit the account per week. Below three per platform per week and the account is starving. At five to eight per platform per week the account has a real chance of consistent scale. Concepts, not variations. Ten variations of the same hook is one concept.
2. Simple account structure
One prospecting campaign per platform. One retargeting campaign per platform. One bottom-funnel or branded-search campaign per platform when the category supports it. Inside each campaign, two to four ad sets max. Inside each ad set, three to five creatives max. Complicated account structures give the platform less data per ad set and lose the auction. Simple is faster to learn from and easier to defend to leadership.
3. Kill rules committed to in advance
Kill rules protect the media buyer from their own bias. Standard rules I have used: on Meta, kill a creative that spends 1x target CPA with no conversion, or a creative with CTR under 0.5% after $50 spent. On TikTok, kill a creative with hold rate under 15% at the six-second mark. On LinkedIn, kill an ad set that spends 3x target CPL with no lead. Committing to kill rules in the flight plan means killing on schedule instead of arguing.
4. Bidding honest to the objective
Cost cap and target CPA bidding work when the account has enough conversion volume. Lowest cost with a spend cap works when it does not. Manual CPM or CPC bidding on newer platforms (Reddit especially) is often necessary to control cost until conversion tracking matures. The mistake is running "target CPA" on an account with 10 conversions a week: the algorithm cannot optimize on that little data.
5. Attribution triangulated on purpose
Platform-reported attribution over-credits itself. Last-click under-credits paid social. Both are wrong most of the time. The working model: platform-reported CPA as a directional signal, self-reported source on the intake or checkout form as an anchor, and a scheduled incrementality test (channel off for two to four weeks) once per quarter as the audit. Media budgets defended by all three survive scrutiny.
First 30 / 60 / 90 days
Days 1 to 30: audit, restructure, prep creative
Pixel and conversion event audit on every platform in the account. Are events firing, are they matching CAPI or the Conversions API. Historical account audit: which campaigns actually produced conversions, which ad sets were vanity. Restructure to the simple account architecture. Land the creative pipeline: brief format, review cadence, production partners (in-house, UGC creator, or studio). Kill rules documented and agreed with the CMO. Deliverable at day 30: clean pixel setup, simplified account structure, creative pipeline live, kill rules committed to. Number moving this phase: creative concepts in the queue for weeks two through six.
Days 31 to 60: first testing wave
First real testing wave ships. Three to five new concepts per platform per week. Every winner tagged and documented. Every loser autopsied briefly (why did it fail: hook, offer, audience, format). Retargeting flows tuned. Landing pages reviewed and, where budget permits, restructured to match the ad creative. First read on which platform is producing signal at what cost. Deliverable at day 60: 24 to 40 new creatives tested per platform, ranked winners identified, ranked losers killed on schedule, first scale plan drafted. Number moving this phase: cost per acquisition versus target on each platform.
Days 61 to 90: scale winners, kill losers, defend the budget
Winning creatives duplicated into scaling ad sets at 20 percent budget lifts every 48 hours. Losing campaigns closed. Retargeting audience quality reviewed (excluded lists, exclusion windows). First incrementality test scheduled or executed on the platform with the biggest budget. Weekly report to the CMO with platform-reported CPA, self-reported intake data, and creative-quality metrics (CTR, hold rate, conversion rate on landing page). Roadmap for months four through twelve with expected creative refresh cadence built in. Deliverable at day 90: scaled winners, killed losers, defended budget, incrementality test result, and a documented creative refresh cadence. Number moving this phase: blended CPA across the paid social program.
Tools I use
Ad platforms. Meta Ads Manager, TikTok Ads Manager, LinkedIn Campaign Manager, Reddit Ads Manager, Pinterest Ads Manager. Native platforms first. Third-party ad management (Smartly, Rockerbox, Skai) only above roughly $200k a month in blended spend.
Conversion tracking. Meta Conversions API, TikTok Events API, LinkedIn Conversions API, server-side GA4 via Google Tag Manager server container. First-party server-side tracking is worth the setup cost on any account above $10k a month.
Creative production. UGC creators sourced through Aspire or direct-book. Studio work for hero and category-education creative. CapCut and Adobe Premiere for editing. Figma for static ad templates.
Creative analytics. Motion (formerly Motion App) or Atria for creative-level reporting across Meta and TikTok. Native reporting for LinkedIn, Reddit, Pinterest.
Attribution. Northbeam or Triple Whale for consumer ecommerce. HubSpot or Salesforce closed-loop attribution for B2B. First-party intake survey on every checkout or lead form for the fraction attribution will always miss.
Incrementality. Meta Lift Studies where budget qualifies. Geo-holdout tests for lower-budget accounts. Scheduled channel-off tests once per quarter as a baseline audit.
What kills the program
1. Not enough creative
Two ads a week, or the same six ads running for three months. The account is starving before the media buyer has done anything wrong. Creative pipeline is the single biggest lever, and most teams under-invest in it.
2. Over-complicated account structure
Twenty-seven ad sets across eleven campaigns because the media buyer wanted "clean segmentation." The platform has too little data per ad set to optimize. Simplify. Fewer, larger ad sets almost always outperform more, smaller ones now.
3. Refusing to kill
Media buyer keeps a losing campaign live because "it will turn around." Kill on schedule. The kill rules exist because bias overrides evidence at the ad-set level. Trust the rules and re-fund learnings into new creative.
4. Chasing platform-reported ROAS
Meta will happily report 5x ROAS on a campaign that is producing 1x incremental revenue. Media plans defended only by platform-reported numbers get audited the moment the CFO asks for a channel-off test. Triangulate from day one.
5. Landing page not matching the ad
Creative wins the click, landing page loses the conversion. If the landing page does not match the ad's promise, angle, and audience, no bid strategy will save the account. Ad-to-page match is a paid social discipline, not just a landing-page team problem.
6. UGC treated as an afterthought
Studio-shot brand creative fills the account. UGC gets one or two spots on the roster. In consumer categories, UGC beats studio on cost-per-outcome most of the time. Treat UGC as a primary creative format, not a supplement.
7. Ignoring platform mismatches
Same brand-voice ad copy on Meta, TikTok, LinkedIn, Reddit, Pinterest. TikTok users tune out ads that read like commercials. Reddit users downvote ads that do not read like the community. Pinterest users search intent-first. Every platform needs its own creative language.
KPIs that matter
Blended CPA versus target. The number the CFO reads first. Track weekly.
Creative concept throughput. Net-new concepts shipped per platform per week. Leading indicator on future performance.
Winner-to-tested ratio. Percentage of tested creatives that hit the scale threshold. Below 15% means the brief is off or the creative team is under-resourced.
Hook rate. On Meta and TikTok, three-second video view rate. Below 25% on TikTok, the creative is failing at the first frame.
Hold rate. Six-second view rate on video. Below 15% on TikTok means the middle of the ad is failing.
Landing page conversion rate. Track by campaign. A drop between the ad CTR and the landing page conversion is the fastest diagnostic tool the paid team has.
Incrementality delta. Quarterly channel-off test result versus platform-reported ROAS. This is the number that keeps the budget defended.
Cost per creative test. Total testing spend divided by concepts tested. Leading indicator on how quickly the account is learning.
FAQ
Which platform should we start with?
Follow the buyer. Consumer under 35 starts on Meta and TikTok. B2B starts on LinkedIn. Consumer over 40 starts on Meta. Pinterest for intent-shopping categories (home, wedding, DIY). Reddit for high-consideration categories where the audience already researches there.
How much budget do we need to test properly?
On Meta and TikTok, roughly $50 to $150 per creative per day for three to five days to get a statistically useful read. LinkedIn needs more because CPMs are higher. Under $3,000 a month is usually not enough to learn on paid social.
How many creatives should we test in a flight?
Three to five per ad set. Fewer and you cannot see angle-versus-angle. More and the budget spreads too thin for any single creative to hit the learning phase.
Should we use broad audiences or interest targeting?
On Meta, broad plus advantage audiences is usually the strongest structure now. Interest targeting still matters on LinkedIn (job title, seniority, company) and TikTok (interest categories, hashtag audiences). Reddit and Pinterest have their own community and keyword targeting.
When should we kill a campaign?
Give it enough spend to exit the learning phase (usually 50 conversions or seven days). If CPA is 2x target after that, kill it. If CTR is under 0.5% on Meta or under 0.8% on TikTok, the creative is broken, not the campaign.
How do we deal with attribution not adding up?
Trust incrementality tests over platform-reported attribution. Turn a channel off for two to four weeks and watch total revenue. Combine with self-reported intake data ("how did you hear about us") on the checkout or lead form. Attribution is a signal, not a verdict.
Do UGC ads really outperform studio ads?
In consumer categories almost always. UGC creative (real people, unpolished, phone-shot) beats studio creative on cost-per-outcome roughly 60 to 80 percent of the time in the accounts I have run. In luxury and B2B, studio creative still wins.
How do we scale a winning campaign without breaking it?
Duplicate the ad set at higher budget rather than editing the winning ad set directly. Increase spend in 20 percent steps every 48 hours, not doubling overnight. Broaden the audience before you broaden the creative.
Related reading
- Paid media that pays back
- Social Search Optimization, the SSO surface explained
- The 19 Ranking Surfaces of Search Everywhere Optimization
- Playbook: Social media strategy
- Playbook: Influencer marketing
If you are trying to make paid social produce defensible revenue, tell me what you are trying to move.
Start a conversation